Abstract XtreemOS is a Linux-based operating system with native support for virtual organizations (VO’s), for building large-scale resource federations. XtreemOS has been designed as a grid operating system, supporting the model of resource sharing among independent administrative domains. We argue, however, that the VO concept can be used to establish either resource sharing or resource isolation, or even both at the same time. We outline XtreemOS ’ fundamental properties and how its native VO support can be used to implement cloud infrastructure and cloud federations. 1 XtreemOS Developing and deploying applications for traditional (single computer) operating systems is well understood. Federated resources like in grid environments, however, are generally perceived as highly complex and difficult to use. The difference lies in the underlying system achitecture. Operating systems provide a well-integrated set of services like processes, files, memory, sockets, user accounts and access rights. Grids, in contrast, add a more or less heterogeneous middleware layer on top of the operating systems of the federated resources. This lack of integration has lead to a lot of complexity, for both users and administrators. To remedy this situation, XtreemOS [7] has been designed as a grid operating system. While being based on Linux, it provides a comprehensive set of services as well as a stable interface for wide-area, dynamic, distributed infrastructures composed of heterogeneous resources spanning multiple administrative domains. The fundamental issues addressed by XtreemOS are scalability and transparency.

Popular web applications like Facebook, Google Search Engine or Amazon have to face major challenges. First of all, the volume of data that has to be managed by these applications goes far beyond the limits of traditional data bases. As a reaction, a new generation of very scalable storing solutions, NoSQL data stores, has been design by the major Internet companies. Another important challenge that popular applications have to deal with is represented by flash crowds. In this context efficient resource provisioning is mandatory in order to avoid unpleasant situations like Amazon crashes from 2008 and 2011. This thesis addresses the problem of resource provisioning in the context of NoSQL data stores. First of all, we analyze the impact of different design decisions from the resource provisioning and we present the necessary steps toward building an accurate model for NoSQL solutions. As a proof of concept, our research presents the performance model of

The same component isolation that made it effective for large distributed telecom systems makes it effective for multicore CPUs and networked applications. By Joe aRmstrong Erlang is a concurrent programming language designed for programming fault-tolerant distributed systems at Ericsson and has been (since 2000) freely available subject to an open-source license. More recently, we’ve seen renewed interest in Erlang, as the Erlang way of programming maps naturally to multicore computers. In it the notion of a process is fundamental, with processes created and managed by the Erlang runtime system, not by the underlying operating system. The individual processes, which are